Here are the times from Saturday October 12th that I was able to collect.
I would like to collect an entire day but our equipment at the moment is not capable of doing so. I have an idea that could make this process better and more accurate which I have posted to our project reddit page. Please post your thoughts there and if there’s enough interest (and if it’s fast enough), we could possibly have this ready to go as early as this or next weekend. There are also plenty of other projects that have been forgotten about. Check those out and if there’s any interest, we can make more things happen.
Last week some people were interested in a comparison time so SDC Men’s A from 2013 was ~51 seconds which would put any time close to 50 seconds as a very competitive freeroll.
Remember, please check out the Project Reddit and tell us what we should work on next.
the Pope says:
wondering how you extract a time from 2 cameras that do not appear to be in sync . i must be missing something obvious.
_______ says:
http://cmubuggy.org/news/2013/10/freeroll-times-fall-56/ second paragraph here says the camera’s are in fact sync’d
Ben says:
Correct, I sync the cameras at the start of the footage before they’re set up on the course.
the \ says:
any idea if how much they drift over the course of a morning?
Ben says:
I try to use the same type of cameras to reduce any chance of this happening which is how these times were taken. I don’t have an easy way right now of checking time drift, but that’s why I’m trying to move to a more standardized camera set. Hopefully we’ll be getting those for next weekend. I can then take the cameras and compare them and we can see how valid these times might be.
Elmo Zoneball says:
If you want to get semi-nerdly about it, you could inject the signal from WWV into the audio track (assuming the cameras have an audio in jack) from a portable short wave receiver at various times, such as between rolls when nothing is happening — especially are the start of the day, and then at the end, then afterward, import the audio tracks from the two cameras into something like Audacity, and compare the time stamps to the audio signals from WWV to see if there’s any drift.
If there IS drift, you could use this technique to measure the correction factor to use to compensate for the drift.
My guess is the drift will be so small as to not matter. Does anyone really need to know the Free Roll time to an accuracy of a hundreth of a second?
Shafeeq says:
Another way would be to record a running stopwatch on both cameras at the start of rolls, and again at the end, and compare with the video’s timestamp.
The spread between contending teams is under a second, so we’d need at least tenths to see anything interesting.
When using multiple stopwatches, the drift at the end of rolls wasn’t measurable, but a stopwatch has just one job to do. Computers are notoriously precise but inaccurate at keeping track of time, and since these cameras are essentially small computers, if their timekeeping is more computer-like than watch-like, they could be a couple seconds off.
You might get a different drift on each day, since the crystal oscillators used in everything vary in frequency with temperature; wristwatches benefit from being the same temperature as your wrist and fancy things like the raceday system either measure the temperature and correct for it, or have heaters to keep themselves at constant temperature.
Ben says:
Great ideas guys,
The stopwatch one is easiest, so I’ll try that this weekend if I remember. Hopefully I remember not to turn off the cameras immediately when I retrieve them.
-Ben